7.2.1.1 MEGA Fabrication and Assembly

MEGA chips shown schematically in Figure 7.1 were constructed from three 100 mm diameter glass wafers and a thin poly(dimethylsiloxane) (PDMS) membrane following a process described by Zeng et al.³¹ Briefly, the valve structure was first lithographically fabricated on the top side of a Borofloat glass wafer. Using backside alignment lithography, the droplet generators were fabricated on the bottom of the same wafer in register with the pump structure. Fluidic and interlayer via holes were drilled and the wafer was thermally bonded to a clean glass substrate at 650°C for 6 h. Pneumatic control manifolds were microfabricated on another glass wafer. Prior to device assembly, the microchannels were coated with 0.1% octadecyltrichlorosilane in dry toluene for 10 min. A 250 μm thick PDMS membrane was sandwiched between a microchip and a pneumatic manifold piece immediately after 2 min UV-ozone activation of the interacting surfaces. A custom Plexiglas manifold was assembled with the MEGA stack to allow oil infusion and droplet collection.

7.2.1.2 Droplet Generation

The carrier oil contains 39.8% (w/w) Dow Corning 5225C Formulation Aid, 30% (w/w) Dow Corning 749 Fluid, 30% (w/w) AR20 Silicone Oil, and 0.2% (w/w) Triton X-100 surfactant. A syringe pump was used to continuously inject carrier oil into the device at various flow rates. The on-chip three-valve diaphragm pump was pneumatically actuated by a homemade solenoid valve system to pump the aqueous solution through the channels. A Welch dry vacuum pump was used to supply constant vacuum (−80 kPa) for valve actuation while the pressure was varied as specified in the text. Pumping was conducted in either four- or three-step fashion under the control of a program written in LabVIEW.

7.2.2.1 MEGA Design and Performance

To enable detection of extremely low frequency events in a vast population, we have developed a series of MEGA systems with 4, 8, 32, and 96 parallel units (Figure 7.1). These multilayer devices consist of a bonded glass–glass microfluidic chip, a PDMS membrane, and a microfabricated manifold wafer. The main fluidic chip contains a pattern of valve seats microfabricated on the top surface, which is connected through a “via” hole to the enclosed microfluidic networks. The assembly is then completed by contact bonding a manifold wafer onto the fluidic chip with a PDMS membrane to form the micropump structure. The microfluidic networks consist of the symmetrically bifurcated channels for introduction of aqueous and oil phases, which form parallel droplet nozzles. Four- and eight-channel MEGAs are based on the single-pump design, but with different droplet generator nozzles: a cross-shaped nozzle in the four-channel MEGA and a T-shaped nozzle in eight-channel devices (Figure 7.1a and b). The four-channel array defines a basic unit that is used to build up multiplexed MEGAs, such as the 32-channel MEGA with eight such arrays integrated onto a 4″ wafer. Further multiplexing of the MEGA on a 4″ wafer is restricted by the number of individual pumps that can be symmetrically arranged in a circle. To achieve a higher density, we designed a new ring micropump composed of three pairs of coaxial ring-shaped valve seats connected by offset channels, as well as corresponding circular displacement trenches (Figure 7.1c). This compact micropump, along with the T-shaped nozzle design, enables the implementation of a 96-channel MEGA on a 4″ wafer. In the 96-channel MEGA, oil channels are connected to the oil inlet holes drilled on the bottom substrate of the device.

For droplet generation, a homemade manifold module is used to infuse oil phase and route generated droplets into PCR tubes, as sketched in Figure 7.2a. The bottom part of the manifold is designed to form a circular oil reservoir when sealed against the MEGA chip. This design yields uniform oil introduction across the whole channel network while minimizing the number of syringe pumps and tubing connections required. The compact ring-shaped micropump in the 96-channel MEGA confers sufficient power to drive multiple droplet generators in parallel. At the same time, the symmetrically designed microfluidic network ensures even fluidic transport, which is crucial for uniform droplet encapsulation. The symmetrical design of the MEGA assures uniform generation of monodisperse droplets in all channels. The microphotograph in Figure 7.2b demonstrates the generation of uniform droplets from a mock PCR mix containing agarose beads by flow shearing at the T-shaped injectors. The mean diameter of the droplets was determined to be 162 μm (2.2 nL) with a deviation of only 3.8% (Figure 7.2c). Such size uniformity is preserved across a range of droplet volumes (1–5 nL; RSD < 5%).

To assess the encapsulation performance of the 96-channel MEGA, single-cell emulsion PCR was performed targeting the KI#128 island on the K12 genome. Bacterial Escherichia coli K12 cells and forward primer modified beads were introduced at a statistical dilution of 0.2 cells and 0.1 beads per 2.5 nL droplet, respectively. After PCR and isolation, flow cytometric analysis (Figure 7.2d) shows that 15% of the total bead population (4423 beads) is strongly fluorescent, corresponding well to a theoretical value of 18% predicted by the Poisson distribution. This good agreement indicates successful single-copy genetic amplification resulted from uniformly distributed cells and beads. A 96-channel MEGA can produce up to 3.4 × 10⁶ droplets per hour (dph), which enables rapid detection of extremely low frequency events in a vast population.

7.2.2.2 Programmable Active Droplet Generation

We have systematically investigated the effects of pulsatile on-chip pumping and flow conditions on droplet formation using the eight-channel MEGA. A three-step actuation sequence was used here to increase the droplet formation rate. We first studied the effects of the oil flow rate, a critical factor in microfluidic droplet generation. Figure 7.3a plots the droplet formation frequency (f_droplet) and droplet size (R_droplet) as a function of the oil flow rate (Q_oil) under an actuation sequence of 40, 50, 40 ms and a pressure of 35 kPa. At the low oil flow rate in regime (a), large uniform droplets were periodically formed, as seen in Figure 7.3b. Interestingly, f_droplet is lower than the pumping frequency (f_pump) of 7.69 Hz and increases along with Q_oil while R_droplet decreases. Microscopic visualization also observed a droplet break-up process, which resembles that of passive cross-flow droplet formation.³⁹ These results indicate that the droplet break-up within regime (a) is predominantly governed by the interaction of viscous shear stresses and interfacial tension between the two immiscible phases and that the force generated by mechanical pump actuation plays a minor role.⁴⁰ As Q_oil was further increased, a narrow transition regime (b) occurred in which droplet generation lacks both periodicity and size uniformity, as seen in Figure 7.3b. A sharp change of droplet generation behavior was observed when f_droplet was tuned to a threshold value of 1.3 μL/min (Figure 7.3a). Within this regime of a wide range of Q_oil, the use of active on-chip pumping enables stable generation of uniform droplets, as exemplified in Figure 7.3b. Several unique phenomena were observed in this regime. First, the droplet generation synchronizes with the pumping frequency, which can be attributed to the pulsatile nature of on-chip pumping that modulates the process of droplet formation. Second, f_droplet is independent of Q_oil, which indicates the dominance of the forces caused by pumping over the viscous shear stresses that depend on the flow parameters. Third, the droplet size can be independently tuned by varying Q_oil while droplet formation remains synchronized with the pumping. Increasing Q_oil leads to higher back pressure that suppresses the pneumatic actuation of the valves and thus reduces the flow rate of the aqueous phase being pumped into the channels.⁴¹ Eventually, the droplet generators will stop working when the back pressure is larger than the pressure that the pump is able to generate (Figure 7.3a). These observations clearly demonstrate that on-chip pumping confers the ability to independently control frequency by adjusting pumping rate and droplet size by varying Q_oil. This unique property is in contrast to the passive droplet generation methods in which droplet generation frequency and droplet size are linked and sensitive to the flow conditions.³³ Figure 7.3 shows that a range of Q_oil needs to be experimentally defined to ensure precisely controlled droplet generation with excellent size uniformity. As expected, the Q_oil range and the droplet sizes were found to be also affected by the pumping conditions, such as f_pump and the pressure for pneumatic actuation.

Figure 7.4a shows that the threshold Q_oil at which droplet generation and pumping start to synchronize shows a roughly linear function of f_pump, especially for the high frequencies. At a fixed pressure (same as in Figure 7.3a) within the range that we commonly used, increasing pumping rate requires a higher oil flow rate to establish stable operation. Since the droplet size is also affected by Q_oil (Figure 7.3a), one should work at oil flow rates away from the boundary conditions such that the adjustment of droplet generation frequency by varying f_pump will require no changes of oil flow and thus have no effects on droplet sizes. In addition to oil flow rate, droplet sizes can be tuned by adjusting the actuation pressure. Droplet volume exhibits a peaking response to increasing actuation pressure under the conditions of pumping and oil flow that we commonly used for fast droplet generation. An example is shown in Figure 7.4b with a pumping sequence of 45, 45, 45 ms and Q_oil at 3 μL/min. The experimental observations in Figures 7.3 and 7.4 provide guidance for exploring proper conditions for independent and precise control of the droplet frequency and sizes.

It is important to note that these droplet generation behaviors are distinct from those of mechanical actuation-based methods that use PDMS or piezoelectric valves.^40,42 In a PDMS valve-based method, the dispersed phase is constantly driven by external pumps and a valve placed close to the droplet generation junction is periodically actuated to alter the instantaneous flow rate of the dispersed phase.⁴¹ It was found that forced droplet generation synchronizes with natural droplet generation only if the actuation frequency is a multiple of the natural frequency, otherwise irregular droplet formation will occur. In our method, synchronized droplet formation can be obtained at any pumping frequency as long as the oil flow rate falls in regime (c) in Figures 7.3 and 7.4. Therefore, our studies suggest a different physical mechanism that should arise from the combination of pulsatile transport and modulation of the dispersed phase, in contrast to the valve-based devices that only modulate the flow being continuously injected. More experimental and theoretical studies are needed to fully elucidate the fluid dynamics of droplet formation driven by pulsing pumping.

Because the on-chip diaphragm pump directly delivers the dispersed phase in our device, the pumping frequency or period is expected to be an important factor in determining the flow rate of the dispersed phase and thus the droplet size. As seen in Figure 7.5a, the droplet size increases and then levels off as the pumping period is extended, consistent with the reported behavior of three-valve micropumps.⁴¹ Adjusting the pumping period provides a means to control droplet generation that is more compatible with computer programming. The top photo in Figure 7.5b shows a stream of two differently sized droplets that was formed by continuously alternating two pumping cycles of 150 and 105 ms. In addition, the droplets can be formed on demand to yield more complex patterns, as demonstrated in the bottom photo of Figure 7.5b. In this case, each repeating sequence consists of seven steps: two consecutive pumping cycles of 130 and 165 ms, 100 ms interval, 110 ms pumping, 300 ms interval, 150 ms pumping, and 500 ms interval. Clearly, the on-chip pumping method allows precise control of droplet generation frequency, size, and timing, which demonstrates its on-demand capability and programmability.

7.3.1.1 Cell Culture

All cell culture and preparation was performed in a class II biosafety cabinet to avoid contamination. Two types of cells, E. coli K12 and nontoxigenic E. coli O157, were grown separately in tryptic soy broth medium overnight at 37°C. E. coli K12 cells were transformed with a 3.9 kb pCR 2.1-TOPO vector to confer ampicillin resistance and grown with 1 mg/mL ampicillin added to the medium. E. coli O157 was untreated and frequently tested for contamination by PCR. Cells were washed three times in 1× PBS and the final cell density was determined using a hemacytometer.

7.3.1.2 Bead and PCR Preparation

Samples were handled in a UV-treated laminar flow hood. Primers specific to the KI#128 island on the K12 genome and the OI#43 island on the O157 genome were designed to prevent cross-amplification between strains. Reverse primers were labeled with 6-carboxyfluorescein (6-FAM) or Cy5 dye on the 5′ end. 5′ Amine-modified forward primers were linked to agarose beads (34 μm mean diameter) via amine-NHS conjugation chemistry. The coupling reaction was performed at a ratio of ~1.5 μmol oligos per gram beads for the pUC18 target. For E. coli cells, equimolar forward primers were used at a concentration of ~0.3 μmol oligo per gram beads. PCR mixes contained forward primer functionalized beads (40 beads/μL) and varied amounts of freshly prepared template DNA or cells.

7.3.1.3 PCR Reaction

The primer sequences specific for E. coli cells are listed in Table 7.1. For E. coli experiments with an average cell concentration of ≤1 cell per droplet, the PCR mix contains 1× AmpliTaq Gold buffer with 1.5 mM MgCl₂, 0.2 mM dNTPs, 0.5 μg/μL heat-inactivated BSA, 0.01% Tween 80, 0.4 μM reverse and 0.04 μM forward primers for each cell target, 0.1 beads/2.5 nL droplet, 0.2 U/μL AmpliTaq Gold polymerase, and various amount of cells. At the average cell concentrations of 10 and 100 cells per droplet (cpd), the PCR mix was optimized to have 3 mM MgCl₂, 2 μM reverse and 0.2 μM forward primers for E. coli K12, with other ingredients kept the same. To minimize DNA/polymerase adsorption on glass and PDMS surfaces, the MEGA device was primed with a coating solution consisting of 1× AmpliTaq Gold buffer with 1.5 mM MgCl₂, 0.5 μg/μL heat-deactivated BSA, and 0.01% Tween 80. Droplets were thermocycled as follows: 10 min at 95°C; six cycles of 30 s at 94°C, 90 s at 65°C, and 60 s at 72°C; eight cycles of 30 s at 94°C, 90 s at 63°C, and 60 s at 72°C; 10 cycles of 30 s at 94°C, 90 s at 60°C, and 60 s at 72°C; 13 cycles of 30 s at 94°C, 90 s at 55°C, and 60 s at 72°C; and 7 min at 72°C.

7.3.1.4 Bead Recovery and Flow Cytometry Quantitation

The emulsion in the PCR tubes was vacuumed to pass a 15 μm mesh filter in a 13 mm diameter plastic filter holder. The droplets were broken and the beads were retained by the filter. After rinsing with isopropanol, 100% ethanol, and 1× Dulbecco’s PBS (DPBS), the beads were recovered using a 5 mL syringe and stored in 1× DPBS for flow cytometry analysis. The bead suspension in PBS was analyzed using a multicolor flow cytometer (FC-500, Beckman-Coulter). FAM-labeled DNA product on the bead surface was quantified against standard beads with different known amounts of fluorophores (Bangs Laboratories Inc., Fishers, IN). The flow cytometry data were analyzed using WinMDI.

7.3.2.1 Digital Single-Cell Pathogen Detection at the Statistically Dilute Regime

We targeted unique genes on the K12 genome (KI#128 island) and the O157 genome (OI#43 island) using primers labeled with different fluorophores. A four-channel MEGA device was operated for ~18 min to obtain ~3000 beads for flow cytometric analysis. A mixed E. coli bacterial sample containing 50% O157 cells was first analyzed at an average cell concentration (C_avg) of 0.2 cpd. As seen in Figure 7.7a, the flow cytometry profile shows four distinct bead populations: 212 FAM-positive beads due to the amplicons from K12 (7.93%, green), 198 Cy5-positive beads for O157 (7.40%, red), 45 double-positive beads due to coexistence of both cell types in a single droplet (1.68%, orange), and 2220 negative beads (82.99%, blue). The O157 cell ratio (O157-positive beads/total positive beads) is then determined to be 0.48, in good accord with the input O157 cell fraction of 0.5.

7.3.2.2 High-Throughput and Sensitive SCGA

The detection of low frequency genetic variations requires high analysis throughput in order to obtain the statistically significant population for the target. The multiplex SCGA process discussed above uses droplets/beads inefficiently because cells are highly dilute so that most droplets are empty. One way to increase the process efficiency is to perform multiplex single-cell genetic analysis at elevated cell density while still keeping the cells of interest statistically dilute. This is realistic because the target pathogen or mutant cells are typically present at very low relative concentrations compared to normal cells. Figure 7.7b demonstrates the use of a four-channel MEGA device to detect E. coli O157 cells at a frequency of 1/10³. By increasing C_avg from 0.2 to 10 cpd, the effective density of O157 cells is raised from 0.0002 to 0.01 cpd, reducing the number of droplets/beads that must be processed by 50-fold. In this case, all beads should be FAM fluorescent due to coencapsulated K12 cells; the presence of O157 cells in the droplets will modify only a fraction of beads with the Cy5 dye. As expected, the vast majority of beads (1846 out of 1961 events, 96.13%) are FAM positive and a small fraction of double-positive beads (102 events, 0.88%) are observed. The measured O157 ratio is determined to be 0.92/10³ (the ratio of double-positive beads over all positive beads divided by C_avg), close to the 1/10³ input. Multiplex SCGA preserves the quantitative performance for pathogen detection even when the average concentration is up to 100 E. coli cells per 2.5 nL droplet. To further improve the detection sensitivity, we assessed the 96-channel MEGA which allows us to encapsulate up to 10⁴ cells within 5 min when operating the device at ~7 Hz per channel. As shown in Figure 7.7c, the experiments at an input ratio of 1/10⁴ at 100 cpd detect 88 and 21 O157-positive events, giving an output fraction of 0.85/10⁴.

To verify that the observed performance is the result of digital quantification of each strain, we compare the percentage of positive beads obtained with various input ratios and average cell concentrations with that predicted by the Poisson distribution (Figure 7.7d). The multiplex detection is seen to follow Poisson statistics even when individual O157 cells were detected within a high background of 100 cpd. The good correspondence indicates successful single-cell emulsion PCR, which allows digital quantification of the absolute cell concentration. For instance, with C_avg = 100 cpd and O157 cells diluted to 1/10⁴ in a K12 background, the O157 cell density is 0.01 cpd and the detection resulted in 0.91% ± 0.04% double-positive beads, consistent with the predicted ratio of 0.995%. Because of the presence of negative events, the average percentage of O157-positive beads is corrected to be 0.93% ± 0.05% (double-positive beads divided by total positive beads), from which the O157 cell concentration is determined to be 3.7 ± 0.2 cells/μL (input 4 cells/μL).

7.3.2.3 Limit of Detection and Dynamic Range

To determine the limit of detection (LOD) of multiplex SCGA for E. coli O157 detection, the assay was carried out at lower pathogenic ratios and C_avg = 100 cpd using a 96-channel MEGA. Figure 7.8a presents a representative analysis at an input ratio of 1/10⁶ where the experimental output ratio (1.4/10⁵) is one order of magnitude higher than that expected. In this case, to define a statistically significant population at the extremely low cell density (0.0001 cpd for O157 cells), a large volume of PCR mix (1500 μL) was used to produce ~8 × 10⁵ droplets within 20 min, which leads to 45,419 total events analyzed and 65 Cy5-positive events detected. The experimental readout at the input of 1/10⁶ (1.8/10⁵ ± 0.66/10⁵, n = 4) is significantly above the expected value (p = .01), indicating that there is significant background signal. It is seen in Figure 7.8a that the peak for K12 cells (FAM-positive only) tails off into the Cy5-positive region, suggesting that the background could be attributed to misamplification caused by the nonspecific binding of O157 primers to the K12 templates. Figure 7.8b is the result obtained with an input ratio of 1/10⁵ that detects 33 Cy5-positive beads. At this pathogenic ratio, the experimental readouts (2.4/10⁵ ± 0.91/10⁵, n = 4) are in the same order as the expected value and significantly different from that obtained at 1/10⁶ at p = .05 using the two-sample Student’s t-test. By subtracting the background, the corrected value at the input ratio of 1/10⁵ is determined to be 0.69/10⁵ ± 0.34/10⁵ (n = 4). Figure 7.8c summarizes the background corrected calibration of E. coli O157 detection as a function of the input O157 fraction, which shows a linear dynamic range for O157 fraction higher than 1/10⁵. From this plot, the concentration detection limit is determined to be 3/10⁵ with a 99% confidence.

7.4.2.1 Cell Culture and Preparation

Cell lines of t(14;18)⁻ TK6 (CRL 8015) and t(14;18)⁺ RL (CRL 2261) lymphoblasts were cultured in RPMI 1640 media supplemented with 10% FBS at 37°C in a 5% CO₂ atmosphere. Cell preparation was performed in a class II biosafety cabinet. Final cell density was determined to be between 2 × 10⁵ and 2 × 10⁶ using a hemacytometer. Cells were washed with PBS, combined with primer-conjugated beads, and mixed with 2% low-melt agarose to a final concentration of 400 beads/μ and 80–120 cells/μ in 1.5% agarose at 40°C.

7.4.2.2 Agarose Droplet Generation

The preparation of the carrier oil and the device operation have been described previously.²⁹ Molten agarose solution mixed with cells and beads was injected into the channels by the on-chip three-valve diaphragm pump. The three valves form the input, displacement, and output of a diaphragm pump and are cycled in a four-step fashion under the control of a LabVIEW graphical interface. The whole MEGA assembly was continually heated using a heated air stream (42°C–45°C) to avoid agarose gelling in the device. Pumping parameters were optimized to produce uniform ~3 nL 1.5% agarose droplets. The sample reservoir was constantly replenished with agarose suspension to sustain droplet generation. Agarose droplets were collected and immediately gelled in 0.5 mL PCR tubes placed in freezer trays.

7.4.2.3 Single-Genome Purification

Agarose droplets were extracted from carrier oil by washing with isopropanol, 100% ethanol, and PBS (10 mL each) and resuspended in 1 mL PBS. To lyse cells, 1 mL 2× SDS lysis buffer (1% SDS, 200 mM EDTA, 20 mM Tris–HCl) and 100 mg/mL proteinase K were added, and samples were incubated overnight at 52°C. Agarose droplets were washed with 2% (w/v) Tween 20 to remove PCR-inhibiting SDS, with 100% ethanol to inactivate any remaining proteinase K, with 0.02% (w/v) Tween 20 five times (10 mL each), and finally 10 mL H₂O. The supernatant was removed prior to emulsion PCR. Samples not immediately used were stored at 4°C in 100% ethanol.

7.4.2.4 Emulsion PCR

The PCR mix contained 1× AmpliTaq Gold buffer with 1.5 mM MgCl₂, 0.2 mM dNTPs, 4 μg/μL heat-inactivated BSA, 0.01% Tween 80, 0.45 μM forward and 0.045 μM reverse primers for each target gene, 0.25 U/μL AmpliTaq Gold polymerase, and 12.5 μL agarose droplets containing purified single-cell genomes in each 50 μL PCR reaction. Agarose droplets were equilibrated in PCR mix in standard 0.5 mL PCR tubes for 30 min at 4°C with occasional agitation. To reemulsify the agarose droplets, 160 μL carrier oil were added and mechanically agitated. Carrier oil was identical to the oil used for initial agarose droplet formation but contained 0.8% (w/w) Triton X-100 surfactant for increased stability. Emulsions were generated by mechanical agitation. Thermal cycling, carried out in a PTC200 thermocycler (MJ Research, Waltham, MA), involved ramping to 95°C (0.1°C/s) in order to melt agarose droplets without merging, followed by a 10 min hot start at 95°C, and 33 cycles of 95°C for 30 s, 60°C for 60 s, 72°C for 90 s, and a final 72°C extension for 5 min. The reactions were cooled to 4°C.

7.4.2.5 Fluorescence Imaging

Spinning disk confocal microscopy was performed with 491 and 561 nm excitation lasers through a 63 × 1.4 NA oil immersion objective (Carl Zeiss, Inc., Thornwood, NY). Images were captured using a Cascade II charge-coupled device. Samples were stained with (50 g/mL) acridine orange or (500 g/mL) propidium iodide. Epifluorescence imaging of amplicon-labeled beads was performed using a 10 × 0.25 NA Nikon Plan objective. Images were contrast adjusted and cropped if needed for optimal viewing.

7.4.2.6 Single-Cell Sequencing

Amplicon-bound beads from emulsion PCR were reamplified in 50 μL TaqMan qPCR reaction in reactions in standard 96-well plates. Beads were counted using a hemocytometer and diluted to approximately one bead per reaction. PCR mix contained 1× Ampli buffer with 5 mM MgCl₂, 0.2 mM dNTPs, 0.025% DMSO, 1 μM ROX reference dye, 0.15 μM forward and 0.15 μM reverse primers and 0.15 μM TaqMan probe for each target gene, and 0.035 U/μL AmpliTaq Gold polymerase. Thermal cycling in an ABI7300 cycler consisted of a 10 min hot start at 95°C, and 20 cycles of 95°C for 15 s, 60°C for 30 s, and 72°C for 30 s. Reactions were loaded on a 1.5% agarose gel and the two amplicons were separated and then excised, purified using a QIAquick Gel Extraction Kit (QIAGEN, Valencia, CA), and sequenced on an ABI3730 at the UC Berkeley Core Sequencing Facility.

7.4.3.1 Single-Cell Encapsulation and Genome Purification

Single lymphoblast cells were encapsulated along with primer-functionalized beads in 1.5% low-melting-point agarose by using a four-channel MEGA. Micropump actuation was optimized to account for the increased viscosity of molten agarose without modification of the microfluidic design. Droplet generation at approximately 40°C for 30 min resulted in the encapsulation of approximately 18,000 cells at up to 0.3 cpd on average. Figure 7.10a shows the generation of uniform 3 nL agarose droplets containing primer-functionalized beads. The inset highlights an example of cell and bead coencapsulation in a single droplet. Encapsulation in agarose droplets enables reproducible single-cell DNA extraction and isolation. Figure 7.10b shows an example of an agarose droplet in which a bead and a cell were coencapsulated (the bead is larger than the cell). Cell lysis and digestion of the DNA-binding histone proteins upon overnight incubation of the gel droplets in SDS lysis buffer containing proteinase K led to complete release of genomic DNA in a cell. The void left by the cell in the agarose was occupied by brightly fluorescent genomic DNA, which exhibited minimal diffusion into the surrounding gel (Figure 7.10c) as a result of the relatively small pore size of 1.5% agarose (~130 nm). An incubation temperature of 52°C facilitated enzymatic protein digestion while preserving the integrity of the agarose droplets. By staining with propidium iodide, we were able to visualize single high-molecular-weight DNA strands protruding from the relatively small agarose droplets. This result indicated that the majority of nuclear proteins were removed by the combination of proteinase K and SDS. The agarose droplets were stable for at least 1 week when stored in ethanol at 4°C, as determined by confocal imaging of DNA diffusion radii.

7.4.3.2 Multiplexed Single-Cell Droplet PCR

A key benefit of agarose encapsulation is the ability to mechanically manipulate the isolated genomic DNA without mixing the genetic content of different cells. The agarose matrix also protects the genomic DNA from physical damage caused by flow shearing during liquid sample preparation. Without going through another microfluidic droplet generation, agarose droplets equilibrated with PCR mix can be easily reemulsified by mechanical agitation in dispersing oil to produce uniform nanoliter-droplet “reactors” for massively parallel single-cell PCR analysis. Excess PCR mix produces microfines (emulsions <1 μm in diameter), which enhance emulsion stability during thermal cycling.⁶⁵ The agarose droplets melt during the hot start phase of PCR and remain liquid throughout the amplification process, maximizing reagent and amplicon diffusion rates. We varied the initial heating rate and tested various concentrations of Triton X-100 as well as combinations of Abil em90 and Span 80 detergents in oil. A slow temperature ramp profile (0.1°C/s) resulted in improved short-term stability, and the addition of BSA (4 mg/mL) to the PCR mix and 0.8% Triton X-100 to the emulsion oil minimized droplet merger over the course of PCR thermal cycling (Figure 7.10d). Fluorescently labeled amplicons bound to the primer-functionalized beads could be seen inside the agarose droplets following PCR (Figure 7.10e). The 34 μm cross-linked beads were selected for their ability to amplify targets exceeding 1 kb in amounts of at least 100 amol per bead.²³ The absence of fluorescence from beads in droplets without cellular genomic DNA indicated that genomic targets were not transferred between agarose droplets.

We demonstrated highly parallel genotyping with single-cell resolution by performing a multiplex PCR assay of cancer cells harboring the t(14;18) translocation at various mutant-to-wild type (RL/TK6) cell ratios. Labeling of the control gene product, β-actin, with the cyanine dye Cy5 enabled the quantification of total cell frequency, whereas the BCL-2/IgH translocation t(14;18) product labeled with FAM spanned BCL-2 and IgH genes across their breakpoint regions and could be used to determine mutation frequency. A representative flow cytometric profile of beads following multiplex PCR amplification with 50% RL cells at an average cell frequency of 0.3 cpd (Figure 7.11a) demonstrated distinct populations of negative beads, Cy5-positive beads with β-actin only, and double-positive beads with FAM-labeled t(14;18) and Cy5-labeled β-actin. Beads containing t(14;18) only were never observed, which further indicates the conservation of single-genome integrity during cell lysis and PCR. By maintaining a constant total cell density in the single-cell regime (0.1–0.3 cpd on average) and varying the relative concentrations of mutant RL and wild-type TK6 cells, we generated a standard curve (Figure 7.11b) to confirm that amplification originated from single cells. In this stochastic regime, the linearity (r = 0.993) of the measured concentration of mutant RL cells with respect to the input in the 0%–100% RL-cell-frequency range tested indicated successful genetic analysis of single cells. Importantly, in the subset of samples further tested, the ratio of total amplicon-positive bead frequency determined by flow cytometry to the microscopically observed cell-encapsulation frequency indicated high PCR efficiency (113% ± 24%).

7.4.3.3 Single-Cell Sequencing

For the single-cell sequencing of both target gene loci, the post-PCR beads were diluted to the stochastic limit and reamplified in a standard 96-well plate, and the products from individual wells were separated by gel electrophoresis. Gel analysis further confirmed amplification from single genomes, and frequencies of β-actin single-positive to double-positive events matched flow cytometry frequency results in the samples tested. Fluorescence-based sorting can be applied to amplicon-labeled beads to sequence only populations of interest and thereby increase efficiency. Size-separated products from single beads were excised from the gel and reamplified for Sanger sequencing. The correct sequences were recovered for both the t(14;18) target and the β-actin control (Figure 7.12). The random nucleotide insertion sequence in t(14;18) matched the unique translocation “fingerprint” determined by sequencing of RL cells in bulk.⁶⁶ The reamplification step bridges single-cell PCR amplification with standard molecular-biology protocols for the sequencing of multiple genetic loci with single-cell resolution at a rate required for meaningful population analysis.

Using this approach, it is possible to sequence multiple target genes from hundreds of single cells in a single experiment and to perform statistical analysis of gene-sequence variation of a cell population. Our agarose cell-encapsulation and emulsion PCR method has the potential to combine with next-generation sequencing technologies to provide additional gains in throughput for single-cell sequencing. In the case of follicular lymphoma, identification of the breakpoint sequence along IgH and BCL-2 genes can help in the elucidation of the mechanisms of erroneous genetic recombination that cause the translocation t(14;18), whereas the random insertion sequence between the two chromosomes is a unique identifier of distinct mutation events.⁶⁶ Alternative single-cell analysis approaches, such as fluorescence in situ hybridization and in situ PCR, are utilized for investigating cell-mutation progression, their throughput is limited, and the inability to recover amplicons prevents the identification of unknown mutations.

Agarose droplet encapsulation offers a robust means for parallel cell lysis and DNA purification of cell types that are difficult to screen using other single-cell PCR techniques. The method can be adapted to encapsulate live single cells for cell culture and subsequent analysis of clonal populations. Although a similar approach was demonstrated recently for the screening and amplification of alginate-encapsulated E. coli colonies containing plasmid libraries,⁶⁷ our approach enables the robust genome purification of mammalian cells, requires 1% of the reagent volume for emulsion PCR, does not require droplet sorting, and maintains single-cell segregation throughout all downstream analyses as a result of the incorporation of primer-functionalized beads as amplicon substrates. Furthermore, single-cell sequencing techniques generally involve cell types that are amenable to lysis during PCR and the sorting of single cells into 96-well PCR plates. However, PCR amplification directly from single cells has been hindered by the lack of a DNA-purification step that would remove histones and other nuclear components that inhibit polymerase activity. Our approach incorporates a highly parallel single-cell-lysis and DNA-purification step, resulting in near 100% amplification efficiency. Finally, the ability to purify genomic DNA from single cells in a supporting matrix opens up the possibility of single-cell epigenetic analysis, such as the methylation-specific PCR of gene regulatory sequences following conversion with bisulfite for sequencing.